Skip to main content
← All episodes
Repeat Purchase

AI in ecommerce: Focus is the Product

· Hosted by Brooks Bostick & Corey Reid

Everyone says AI means doing more. Brooks and Corey argue the opposite: the best way for ecommerce brands to use AI is to find focus, pick one problem and move faster on it.

The episode opens with a public service announcement about apps and vibe-coded features on your storefront, then gets into how to cut through the AI noise. Corey shares how he uses AI as an analyst to get from raw data to a decision, and Brooks explains why the most successful brands they work with do one thing really well.

In this episode

  • A PSA on apps and vibe coding: build things yourself, but know what your apps and features touch, and never expose API keys on the front end of your Shopify store
  • Just because you can doesn’t mean you should: start with one step that adds value instead of chasing every AI use case
  • AI as your analyst: how Corey uses AI to build subscription cohort analysis and find where customers drop off, so you get to the takeaway faster
  • A takeaway you can run this week: find your biggest churn point, such as between the second and third order, and test one change, like a free gift in that order
  • Activity vs. achievement: nobody gets paid for the number of product pages they make; they get paid when customers buy and come back
  • Ecommerce is an endurance sport: why the most successful brands nail one thing instead of doing ninety
  • Which AI tools to use: Shopify Sidekick for flows and custom reports, Claude for deeper analysis, and why Corey gives AI read access but rarely write access

Repeat Purchase is a podcast by on/Sight about growing ecommerce brands, with practical episodes for Shopify merchants.

Full transcript

Read the full transcript

Brooks Bostick 00:00
Welcome to Repeat Purchase. Before we start, the fine print.

Corey Reid 00:04
Results may vary. Your conversion rate is your own. Past performance of a Shopify store is not indicative of future revenue. Side effects of listening may include a longer to-do list.

Brooks Bostick 00:11
And a better night's sleep.

Corey Reid 00:12
I'm Corey, and I lead growth for our clients at on/Sight.

Brooks Bostick 00:16
And I'm Brooks. I run on/Sight, a Shopify agency that works with CPG and direct-to-consumer brands.

Corey Reid 00:21
We spend our days inside stores people love, figuring out what's working, what's broken and what to do next.

Brooks Bostick 00:27
So every episode, we take one ecommerce topic and get through it in fifteen minutes.

Corey Reid 00:31
Site, email, ads, subscriptions, merchandising. Whatever's keeping a founder up at night.

Brooks Bostick 00:37
Now, the name. Yes, we're a little bit obsessed with the second order.

Corey Reid 00:41
Because for a DTC brand, retention is the whole game. Acquiring a customer is expensive, and keeping one is where the business actually gets built.

Brooks Bostick 00:49
But it's also a promise. If we do this right, you'll be back.

Corey Reid 00:53
No slides, no jargon, no forty-minute detour.

Brooks Bostick 00:57
One question, one honest answer, and probably one argument.

Corey Reid 01:02
Probably about subscriptions.

Brooks Bostick 01:04
Yeah, definitely about subscriptions.

Corey Reid 01:06
Well, let's get into it.

Brooks Bostick 01:08
Hey, Corey.

Corey Reid 01:09
What's up, dude?

Brooks Bostick 01:10
How's it going?

Corey Reid 01:11
Well, I'm back home after a very long weekend of travel through three cancellations, and I'm feeling great.

Brooks Bostick 01:20
Somebody pulled the router on your flight, huh?

Corey Reid 01:23
That darn Arkansas router really got me.

Brooks Bostick 01:27
I tell you, those routers sure do mess you up.

Okay, let's get to our actual topic. This grew out of a public service announcement we recorded a couple of episodes ago that got cut for time, but I think it's really important, and it's come up again.

There's a lot of weird stuff going on in the tech world, and ecommerce is very tech-driven. So let me start with the original PSA, because it's very relevant. If you have apps, or you're vibe coding something on your ecommerce site, on your storefront, on the way your business makes money and survives, and you don't know what those things do, please take the time to understand them, or pay somebody to help you understand them.

There's so much vulnerability in the tech world right now. Some of our largest banks and some of our biggest tools have been hit, with problems spreading to CNN and other giant websites and organizations. It's a very vulnerable time to be in the digital world. I'm a huge advocate for building things yourself, trying stuff, vibe coding and taking a lot of steps on your own. But if you're doing those things, make sure you understand the impact of what you're doing and what your app or features touch. You can easily expose API keys on the front end of Shopify, and that can cause huge issues for your business. That happened a couple of weeks ago with a number of different Shopify apps. So I just want to call that out. Yes, I think building things is amazing, but please make sure you know what you're doing.

Corey Reid 03:14
Wow, you said that well.

Brooks Bostick 03:16
Thank you. Okay, so on that note, let's talk about AI for ecommerce. I think this topic will span multiple episodes, and it's always changing. But first and foremost, take a deep breath. We're all overwhelmed by AI. Even heavy users like us find it really hard to stay at the forefront. So, Corey, since AI is so overwhelming, how do we sift through all the noise and figure out the right way to use it to benefit ecommerce?

Corey Reid 03:54
We use a phrase a lot at on/Sight: just because we can do anything doesn't mean we should do everything. Applied to AI, I think it's less about finding where the most fruit is going to come from, and much more about taking a single step forward to find value for your business.

There are lots of directions someone could go: private applications, vibe coding, really deep work on the CRM. My view is that you should start with the things that used to take a lot of manual human time, like data analysis, and ask how you can speed them up so you get to the answer faster. Because the goal of data analysis isn't to analyze data. The goal is to come away with a takeaway you can act on.

So I want to steer us toward using AI to make the things that used to take a long time move quickly, so we can respond to the data faster. Here's an example. I used to build cohort analyses for the ecommerce businesses I worked for, especially subscription businesses trying to drive high LTV, where we needed to understand CAC-to-LTV ratios and all of that. Those used to be so painful to build. I know there are a thousand tools out there, and you can do it in Recharge, but you often get a very narrow view: LTV for subscribers and their subscription orders only. I want to see all the one-time orders too.

Lately I've been using AI to take these huge swaths of data and cut them so I can understand how our customers are ordering, when they're dropping off, when they're leaving their subscriptions, and what we could do better to keep them longer. I'm using AI like an analyst, cutting through spreadsheets to help me get to the right answers. I'm finding that very beneficial. It isn't some massive setup with tons of agents and Claude Code open all the time. It's simply taking data and analyzing it better, and it's been hugely impactful for me.

Brooks Bostick 06:27
That's a little like finding the signal in the noise. There's a lot of noise with AI, and in ecommerce especially, because we live in the digital world, there's so much you can do. But doing more without understanding what you're actually driving toward doesn't help you at all. So what you're saying is, you need to set the foundation: use AI to understand large amounts of data, then come up with a direction for your business. That's the most impactful use of AI you see right now. And we might not be talking directly to ecommerce engineers here, but we are talking to the people who steer ecommerce businesses and are responsible for their performance.

Corey Reid 07:21
Yeah, and I can give a really quick takeaway on this one. Run a cohort analysis for your subscription business. Find where your biggest drop-off of customers is. Churn. Let's say it's between the second and third order. Do one thing you can measure to see whether you can make that drop-off better, so fewer people leave. By the law of large numbers, if you can change the behavior of those customers, they'll probably stick around longer, and you'll help your entire LTV. Maybe it's putting a free gift in that order and seeing whether those customers stay longer. Simply figure out where the biggest gap is, then go run.

Brooks Bostick 08:09
I love that. So it's not necessarily doing more. In some cases it is, but it's figuring out how to use AI to do better. And it's not really positioned that way. If you read the news or go on LinkedIn, although don't do that too much, every platform is saying more, more, more.

Corey Reid 08:32
I think a lot of it is valuing activity over achievement. Nobody gets paid by the number of product pages they make. You get paid by the number of customers who buy your product and keep coming back. So it's about getting back to the root of the job you're trying to do and how you can do it well. Taking a step back and asking, can I add one percent more value today than I did yesterday?

Brooks Bostick 09:00
And the pace it moves at is overwhelming. I think that's the business model for AI. But we get to choose how we use it and what's going to be most impactful for us, because it varies from person to person. You go on YouTube and find some AI "thought leader," and they're applying it to their own specific use case, which might not apply to you. If we have the ability to do everything, what are the things we can use it for that actually make an impact?

I think that's a great way to position it, and I can feel it as we talk: it makes AI feel less overwhelming, because now I'm looking at it as a way to help me focus rather than a way to do more. More comes with focus, but you have to have focus first. You can't just lean into AI and say, cool, I'm going to create seven new themes tomorrow and fundamentally change the business. You might grow revenue by zero, or you might actually hurt the business. Without that focus, we can't use the tool to its full potential.

Corey Reid 10:15
I feel like a lot of our conversations come back to focus, which is funny, because I have ADHD and can't focus at all.

Brooks Bostick 10:27
They're all aspirational conversations.

Corey Reid 10:29
Aspirational. But in a distracted, overwhelming environment, sometimes the best thing we can do is give ourselves a little peace and focus and ask, what's the job I'm trying to do? If you're an ecommerce manager, it's making more money for the company. Sometimes the best way to do that is to pick a spot, run toward it, measure it, then pick another spot, run toward it and measure it. Don't get overwhelmed by everything you could be doing, because that just creates a lot of churn in your mind. It's not beneficial.

Brooks Bostick 11:03
That's the social media world we live in, too. We should always be doing more, differently, better, more beautifully. Maybe we should just be doing one thing really well. Look at the successful businesses we work with: that's their method. That's how they became successful. It isn't, "We're going to do ninety things really well." It's, "We have this one thing, and we're going to kill it." We've seen that over and over again. It's almost an industry secret, because now everybody is pushing toward diversification and doing as much as possible. But if you look at the really successful businesses out there, especially in ecommerce, they're nailing one thing and doing it really, really well. That's a good benchmark too.

It still takes time to grow a business. We see a lot of overnight successes in ecommerce, especially in CPG, but that doesn't happen to everybody. Ecommerce is still very much an endurance sport, and that's okay. It can take years. on/Sight has been around for eleven years, and we've done a ton of great work. I could look at the other businesses we compete against and say we're not there, we're not good enough. But at the same time, this is our path and our trajectory. As long as we're doing everything we can to reach our goal, that's enough. And AI is telling us the complete opposite story.

I want to round this out with a question about AI tooling, because I want to come back to this topic. Shopify has Sidekick, which I think is really cool, and they push a lot of people to it. Is that a tool you'd recommend for ecommerce businesses? Or are we looking at a hybrid of Sidekick and ChatGPT? Where do all these tools fit, and what can I do today to start getting that focus?

Corey Reid 13:02
I use Sidekick when I'm creating flows or doing something on the Shopify side. I use it much less for data analysis. I'm pretty much a Claude believer. I like how they've built their product, and their tools are really focused on enterprise. I also tend to download data and give it to Claude rather than giving it full access. I barely give it any write access. I want it to be able to read things and find opportunities, but I don't want it making a bunch of changes. I want to stay in the driver's seat. So I'm doing pretty much everything through Claude at this point.

Brooks Bostick 13:48
I'll add that Shopify is pushing Sidekick heavily, and its capabilities seem to grow week over week and month over month. Building custom reports with Sidekick is exceptional. Custom reporting in Shopify used to be very manual. You had to write code, or understand code well enough, to get exactly what you wanted. With Sidekick, you can just ask in plain language: "Give me my mobile conversion rate over the past two weeks in a report," and it builds it for you. Then you can take that report and slice and dice it further with a tool like Claude. That's a great flow for most people. So there are different ways to get to a solid result.

Okay, I think that's enough for the first episode on this. We've already covered a million different things, and we'll need some follow-ups. I know it's overwhelming. AI is in a strange place, and it's moving faster than anybody could have imagined. But like you said, Corey, focus is the product. Focus is what we're all trying to achieve.

Corey Reid 15:00
Well said. Who said that? Who said focus is the product?

Brooks Bostick 15:03
I think I did.

Corey Reid 15:05
I think I wrote that in a deck. All right, Brooks, I appreciate the conversation. It felt a little like a therapy session, which was great.

Brooks Bostick 15:10
Thanks, Corey. Ecommerce isn't easy, but we're here for a reason.

Corey Reid 15:17
All right, man. Have a great day.

Brooks Bostick 15:19
And that's Repeat Purchase. Subscribe, review, and send us your questions.

Corey Reid 15:23
One question, fifteen minutes, every week.

Brooks Bostick 15:26
And remember, you paid nothing for this.

Corey Reid 15:29
Go get that next win.

Brooks Bostick 15:31
And come back next week. We'll be here.

Let’s Talk

Whether you've got a project in mind, a wild idea, or just want to talk shop—we're always up for a good convo. Drop us a note and let's see what we can build and grow together.